This proposal is part of a genome-wide association study (GWAS) to detect susceptibility genes for schizophrenia using an adaptive multistage design that controls the false discovery rate. Stage 1 of this study involves genotyping 665,000 SNPs in 750 schizophrenia cases from the CATIE study and 750 group matched controls from the NIMH repository. This part will be completed soon via an agreement between UNC-CH, CATIE, Eli Lilly, and Perlegen. In this project we propose Stages 2 and 3. In Stage 2, we propose to follow up 16,000 SNPs from Stage 1 GWAS in 5,000 cases and controls from NIMH repository to identify common causal variants for schizophrenia while controlling the false discovery rate at the 0.10 level. Stage 2 SNPs will be selected using 1) a pure statistical approach ensuring that about 80% of the SNPs with "interesting" effects are selected for the follow up study, 2) a semi-automated bioinformatic approach to integrate CATIE WGA data with other sources of information (other GWAS, CNP, linkage, microarray, mouse models, pathway information, literature) to include genes/regions initially that are supported by multiple lines of evidence, and 3) a two-day meeting with all participants plus consultants to decide on the final selection of SNPs. After completing the genoptyping and analyses, we will perform a "validation" Stage 3 study by genotyping 300 SNPs in 5,500 individuals from 1,400 families. The goals are to verify that the common variants identified are not the result of population stratification or ascertainment artifacts and study possible population differences. Finally, explicitly secondary analyses will be performed to search for genotype-genotype and genotype-environment interactions, genetic effects on neurocognitive measures/drug-response, and use our artificial intelligent "model discovery" tool to search for schizophrenia subtypes. All genotype and clinical data will be deposited in the public domain through the NIMH repository. Although no single GWA study can prove or refute the common disease/common variant (CDCV) model for schizophrenia, we aspire to be one of the fundamental studies in support of this goal. If CDCV genetic variation is detected at a "proof beyond a reasonable doubt" level, then this has the potential to lead to revolutionary changes in psychiatry. If CDCV is not found, then the genetic dissection of schizophrenia will require the next generation of sequencing technologies, which may be on the five year horizon. If we are successful in conclusively identifying genetic variation that confers risk for schizophrenia, the public health ramifications could be enormous. Case-finding efforts could be greatly improved and we could learn far more about how this devastating disease develops. [unreadable] [unreadable] [unreadable] [unreadable]